of course, this general topic is off-topic as it is not directly relatod to the Boost Graph Library. And I want to apologize for that in the first place. However, I think that many using the BGL may also have to deal sometimes with such problems. Therefore, I am addressing this question here and hope to find some answers ;) The question is about permuting a graph in order to assess the statistical significance of a certain aspect of the original graph. Sometimes, it is enough to randomly re-wire the existing edges, but other times, more elaborate techniques are necessary.
I don't know of any specific literature on this topic. My knowledge on statistical aspects of graphs stops after the first measurement :) Sorry I can't be of more help. Andrew